feat: apply new helm structure

use minio s3 for savepoint and checkpoint path
separate task-manager, job-manager and operator
use statefulset for task-manager to handle replication
support basic credential for download jar request
update to flink 1.20.1
This commit is contained in:
2025-04-05 01:39:02 +03:30
parent 7f78faeed7
commit 830e265162
26 changed files with 386 additions and 256 deletions

View File

@@ -0,0 +1,33 @@
{{- define "flink.env" -}}
- name: JOB_MANAGER_RPC_ADDRESS
value: "localhost"
- name: NAMESPACE
value: {{ .Release.Namespace }}
- name: FLINK_PROPERTIES
value: |
jobmanager.rpc.address: {{ .Release.Name }}-flink-job-manager
jobmanager.memory.process.size: {{ .Values.flink.jobManager.processMemory }}
taskmanager.memory.process.size: {{ .Values.flink.taskManager.processMemory }}
taskmanager.data.port: 6125
taskmanager.numberOfTaskSlots: {{ .Values.flink.taskManager.numberOfTaskSlots }}
parallelism.default: {{ .Values.flink.parallelism.default }}
state.backend: {{ .Values.flink.state.backend }}
rest.port: 8081
rootLogger.level = DEBUG
rootLogger.appenderRef.console.ref = ConsoleAppender
high-availability.type: kubernetes
kubernetes.namespace: {{ .Release.Namespace }}
kubernetes.cluster-id: {{ .Values.clusterId | default (print .Release.Name "-cluster") }}
execution.checkpointing.interval: {{ .Values.flink.checkpoint.interval }}
execution.checkpointing.mode: {{ .Values.flink.checkpoint.mode }}
state.checkpoints.dir: s3://{{ .Release.Name }}-minio:9000/checkpoints
state.backend.rocksdb.localdir: /opt/flink/rocksdb
high-availability.storageDir: /opt/flink/ha
state.savepoints.dir: s3://{{ .Release.Name }}-minio:9000/savepoints
state.backend.incremental: {{ .Values.flink.state.incremental }}
rest.profiling.enabled: true
s3.endpoint: http://{{ .Release.Name }}-minio:9000 # Use Kubernetes service name
s3.path.style.access: true
s3.fs.hadoop.impl: org.apache.hadoop.fs.s3a.S3AFileSystem # Keep for compatibility
fs.s3a.aws.credentials.provider: com.amazonaws.auth.DefaultAWSCredentialsProviderChain
{{- end }}

View File

@@ -1,10 +0,0 @@
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: {{ .Values.flink.state.data.pvcName }}
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: {{ .Values.flink.state.data.size }} # Use size defined in values.yaml

View File

@@ -1,165 +0,0 @@
{{- define "flink.env" -}}
- name: JOB_MANAGER_RPC_ADDRESS
value: "localhost"
- name: NAMESPACE
value: {{ .Release.Namespace }}
- name: FLINK_PROPERTIES
value: |
jobmanager.rpc.address: localhost
jobmanager.memory.process.size: {{ .Values.flink.jobManager.processMemory }}
taskmanager.memory.process.size: {{ .Values.flink.taskManager.processMemory }}
taskmanager.data.port: 6125
taskmanager.numberOfTaskSlots: {{ .Values.flink.taskManager.numberOfTaskSlots }}
parallelism.default: {{ .Values.flink.parallelism.default }}
state.backend: {{ .Values.flink.state.backend }}
rest.port: 8081
rootLogger.level = DEBUG
rootLogger.appenderRef.console.ref = ConsoleAppender
high-availability.type: kubernetes
kubernetes.namespace: {{ .Release.Namespace }}
kubernetes.cluster-id: {{ .Values.clusterId | default (print .Release.Name "-cluster") }}
execution.checkpointing.interval: {{ .Values.flink.checkpoint.interval }}
execution.checkpointing.mode: {{ .Values.flink.checkpoint.mode }}
web.upload.dir: {{ .Values.flink.state.data.dir }}/web-upload
state.checkpoints.dir: file://{{ .Values.flink.state.data.dir }}/checkpoints
state.backend.rocksdb.localdir: file://{{ .Values.flink.state.data.dir }}/rocksdb
high-availability.storageDir: file://{{ .Values.flink.state.ha.dir }}
state.savepoints.dir: file://{{ .Values.flink.state.savepoints.dir }}
state.backend.incremental: {{ .Values.flink.state.incremental }}
rest.profiling.enabled: true
{{- end }}
{{- define "flink.volumeMounts" -}}
- name: flink-data
mountPath: {{ .Values.flink.state.data.dir }}/data
- name: flink-data
mountPath: {{ .Values.flink.state.data.dir }}/rocksdb
subPath: rocksdb
- name: flink-data
mountPath: {{ .Values.flink.state.data.dir }}/checkpoints
subPath: checkpoints
- name: flink-data
mountPath: {{ .Values.flink.state.data.dir }}/web-upload
subPath: web-upload
- name: flink-ha
mountPath: {{ .Values.flink.state.ha.dir }}
- name: flink-savepoints
mountPath: {{ .Values.flink.state.savepoints.dir }}
{{- end }}
{{- define "flink.volumes" -}}
- name: flink-data
persistentVolumeClaim:
claimName: {{ .Values.flink.state.data.pvcName }}
- name: flink-savepoints
persistentVolumeClaim:
claimName: {{ .Values.flink.state.savepoints.pvcName }}
- name: flink-ha
persistentVolumeClaim:
claimName: {{ .Values.flink.state.ha.pvcName }}
{{- end }}
apiVersion: apps/v1
kind: Deployment
metadata:
name: {{ .Release.Name }}-flink
labels:
app.kubernetes.io/name: {{ .Release.Name }}-flink
app.kubernetes.io/instance: {{ .Release.Name }}
spec:
replicas: 1
strategy:
type: Recreate
selector:
matchLabels:
app.kubernetes.io/name: {{ .Release.Name }}-flink
app.kubernetes.io/instance: {{ .Release.Name }}
template:
metadata:
labels:
app.kubernetes.io/name: {{ .Release.Name }}-flink
app.kubernetes.io/instance: {{ .Release.Name }}
spec:
serviceAccountName: {{ include "flink-kube-operator.serviceAccountName" . }}
initContainers:
- name: volume-mount-hack
image: {{ .Values.flink.image.repository }}:{{ .Values.flink.image.tag }}
runAsUser: 0
command: ["sh", "-c", "chown -R flink {{ .Values.flink.state.data.dir }}/data {{ .Values.flink.state.data.dir }}/rocksdb {{ .Values.flink.state.data.dir }}/checkpoints {{ .Values.flink.state.data.dir }}/web-upload {{ .Values.flink.state.ha.dir }} {{ .Values.flink.state.savepoints.dir }}"]
volumeMounts:
{{- include "flink.volumeMounts" . | nindent 12 }}
containers:
- name: jobmanager
image: {{ .Values.flink.image.repository }}:{{ .Values.flink.image.tag }}
imagePullPolicy: Always
args: ["jobmanager"]
ports:
- containerPort: 6123 # JobManager RPC port
name: rpc
- containerPort: 6124 # JobManager blob server port
name: blob
- containerPort: 6125 # JobManager queryable state port
name: query
- containerPort: 8081 # JobManager Web UI port
name: ui
env:
{{- include "flink.env" . | nindent 12 }}
- name: POD_IP
valueFrom:
fieldRef:
fieldPath: status.podIP
volumeMounts:
{{- include "flink.volumeMounts" . | nindent 12 }}
- name: taskmanager
image: {{ .Values.flink.image.repository }}:{{ .Values.flink.image.tag }}
imagePullPolicy: Always
args: ["taskmanager"]
ports:
- containerPort: 6121 # TaskManager data port
name: data
- containerPort: 6122 # TaskManager RPC port
name: rpc
env:
{{- include "flink.env" . | nindent 12 }}
- name: POD_IP
valueFrom:
fieldRef:
fieldPath: status.podIP
volumeMounts:
{{- include "flink.volumeMounts" . | nindent 12 }}
- name: operator
securityContext:
{{- toYaml .Values.securityContext | nindent 12 }}
image: "{{ .Values.image.repository }}:{{ .Values.image.tag | default .Chart.AppVersion }}"
imagePullPolicy: {{ .Values.image.pullPolicy }}
ports:
- name: http
containerPort: {{ .Values.service.port }}
protocol: TCP
env:
- name: FLINK_API_URL
value: localhost:8081
- name: SAVEPOINT_PATH
value: file://{{ .Values.flink.state.savepoints.dir }}
- name: NAMESPACE
value: "{{ .Release.Namespace }}"
resources:
{{- toYaml .Values.resources | nindent 12 }}
volumeMounts:
{{- include "flink.volumeMounts" . | nindent 12 }}
volumes:
{{- include "flink.volumes" . | nindent 8 }}
{{- with .Values.nodeSelector }}
nodeSelector:
{{- toYaml . | nindent 8 }}
{{- end }}
{{- with .Values.affinity }}
affinity:
{{- toYaml . | nindent 8 }}
{{- end }}
{{- with .Values.tolerations }}
tolerations:
{{- toYaml . | nindent 8 }}
{{- end }}

View File

@@ -1,7 +1,7 @@
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: {{ .Values.flink.state.ha.pvcName }}
name: {{ .Release.Name }}-{{ .Values.flink.state.ha.pvcName }}
spec:
accessModes:
- ReadWriteOnce

View File

@@ -0,0 +1,84 @@
apiVersion: apps/v1
kind: Deployment
metadata:
name: {{ .Release.Name }}-flink-job-manager
labels:
app: {{ .Release.Name }}-flink-operator
component: {{ .Release.Name }}-flink-job-manager
spec:
replicas: 1
strategy:
type: Recreate
selector:
matchLabels:
app: {{ .Release.Name }}-flink-operator
component: {{ .Release.Name }}-flink-job-manager
template:
metadata:
labels:
app: {{ .Release.Name }}-flink-operator
component: {{ .Release.Name }}-flink-job-manager
spec:
serviceAccountName: {{ include "flink-kube-operator.serviceAccountName" . }}
initContainers:
- name: volume-mount-hack
image: {{ .Values.flink.image.repository }}:{{ .Values.flink.image.tag }}
runAsUser: 0
command: ["sh", "-c", "chown -R flink {{ .Values.flink.state.ha.dir }}"]
volumeMounts:
- name: flink-ha
mountPath: {{ .Values.flink.state.ha.dir }}
containers:
- name: jobmanager
image: {{ .Values.flink.image.repository }}:{{ .Values.flink.image.tag }}
imagePullPolicy: Always
args: ["jobmanager"]
ports:
- containerPort: 6123 # JobManager RPC port
name: rpc
- containerPort: 6124 # JobManager blob server port
name: blob
- containerPort: 6125 # JobManager queryable state port
name: query
- containerPort: 8081 # JobManager Web UI port
name: ui
env:
{{- include "flink.env" . | nindent 12 }}
- name: POD_IP
valueFrom:
fieldRef:
fieldPath: status.podIP
- name: S3_ENDPOINT
value: "http://minio-service:9000"
- name: AWS_ACCESS_KEY_ID
valueFrom:
secretKeyRef:
name: {{ .Release.Name }}-flink-secrets
key: minio_access_key
- name: AWS_SECRET_ACCESS_KEY
valueFrom:
secretKeyRef:
name: {{ .Release.Name }}-flink-secrets
key: minio_secret_key
volumeMounts:
- name: flink-ha
mountPath: {{ .Values.flink.state.ha.dir }}
volumes:
- name: flink-ha
persistentVolumeClaim:
claimName: {{ .Release.Name }}-{{ .Values.flink.state.ha.pvcName }}
{{- with .Values.nodeSelector }}
nodeSelector:
{{- toYaml . | nindent 8 }}
{{- end }}
{{- with .Values.affinity }}
affinity:
{{- toYaml . | nindent 8 }}
{{- end }}
{{- with .Values.tolerations }}
tolerations:
{{- toYaml . | nindent 8 }}
{{- end }}

View File

@@ -0,0 +1,28 @@
apiVersion: v1
kind: Service
metadata:
name: {{ .Release.Name }}-flink-job-manager
labels:
app.kubernetes.io/name: {{ .Release.Name }}-flink-job-manager
app.kubernetes.io/instance: {{ .Release.Name }}
spec:
ports:
- name: flink-web-ui
port: 8081
targetPort: 8081
- name: rpc
port: 6123
targetPort: 6123
- name: blob
port: 6124
targetPort: 6124
- name: query
port: 6125
targetPort: 6125
- name: operator
port: 3000
targetPort: 3000
selector:
app: {{ .Release.Name }}-flink-operator
component: {{ .Release.Name }}-flink-job-manager
type: ClusterIP # Change to LoadBalancer if you want external access

View File

@@ -1,10 +0,0 @@
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: {{ .Values.flink.state.savepoints.pvcName }}
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: {{ .Values.flink.state.savepoints.size }} # Use size defined in values.yaml

View File

@@ -1,19 +0,0 @@
apiVersion: v1
kind: Service
metadata:
name: flink
labels:
app.kubernetes.io/name: {{ .Release.Name }}-flink
app.kubernetes.io/instance: {{ .Release.Name }}
spec:
ports:
- port: 8081
name: flink-web-ui
targetPort: 8081
- port: 3000
name: operator
targetPort: 3000
selector:
app.kubernetes.io/name: {{ .Release.Name }}-flink
app.kubernetes.io/instance: {{ .Release.Name }}
type: ClusterIP # Change to LoadBalancer if you want external access

View File

@@ -0,0 +1,58 @@
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: {{ .Release.Name }}-flink-task-manager
labels:
app: {{ .Release.Name }}-flink-operator
component: taskmanager
spec:
serviceName: {{ .Release.Name }}-flink-task-manager
replicas: {{ .Values.flink.taskManager.replicas }}
selector:
matchLabels:
app: {{ .Release.Name }}-flink-operator
component: {{ .Release.Name }}-flink-task-manager
template:
metadata:
labels:
app: {{ .Release.Name }}-flink-operator
component: {{ .Release.Name }}-flink-task-manager
spec:
serviceAccountName: {{ include "flink-kube-operator.serviceAccountName" . }}
containers:
- name: task-manager
image: {{ .Values.flink.image.repository }}:{{ .Values.flink.image.tag }}
imagePullPolicy: Always
args: ["taskmanager"]
env:
{{- include "flink.env" . | nindent 8 }}
- name: POD_IP
valueFrom:
fieldRef:
fieldPath: status.podIP
- name: S3_ENDPOINT
value: "http://minio-service:9000"
- name: AWS_ACCESS_KEY_ID
valueFrom:
secretKeyRef:
name: {{ .Release.Name }}-flink-secrets
key: minio_access_key
- name: AWS_SECRET_ACCESS_KEY
valueFrom:
secretKeyRef:
name: {{ .Release.Name }}-flink-secrets
key: minio_secret_key
volumeMounts:
- name: rocksdb-storage
mountPath: /opt/flink/rocksdb
resources:
{{- toYaml .Values.flink.taskManager.resources | nindent 10 }}
volumeClaimTemplates:
- metadata:
name: rocksdb-storage
spec:
accessModes: [ ReadWriteOnce ]
resources:
requests:
storage: {{ .Values.flink.taskManager.storage.rocksDb.size }}